The AI tool called the Cryosection Histopathology Assessment and Review Machine (CHARM) studies images to quickly pick out the genetic profile of a kind of tumor called glioma, a process that currently takes days or weeks, said Kun-Hsing Yu, senior author of a report released Friday in the journal Med. Surgeons use detailed diagnoses to guide them while they operate, Yu said, and the ability to get them rapidly could improve patients’ outcomes and spare them from multiple surgeries.
The tool uses machine learning algorithms to analyze images of brain tumor samples obtained during surgery. Researchers trained CHARM by comparing its analyses with the diagnoses of patients. The system was able to match or outperform other AI systems in identifying the genetic profile of the tumors. While it may not be as accurate as current genetic tests, CHARM can predict a tumor’s profile almost instantly. This rapid analysis could enable doctors to proceed with appropriate treatment without the delay of scheduling and performing additional surgeries.
CHARM can distinguish between malignant and benign tumor cells and determine the tumor’s grade, which measures its aggressiveness. These determinations are made by human pathologists during surgery, but CHARM could eliminate the need for a 10-to-15-minute wait or the presence of a pathologist during the operation.
The study of demonstrated promising results, it is important to test CHARM in real-world settings, as stated by the researchers. This work is part of a broader effort to utilize artificial intelligence in the field of cancer diagnosis and treatment. Some AI systems have shown the ability to accurately identify individuals at high risk of pancreatic, lung, and breast cancer.
